Learning how critical cell regulatory pathways are controlled may lead to new opportunities for cancer treatment. We recently identified the transcription factor Pokemon as a central regulator of the important tumor suppressor ARF. Pokemon is overexpressed in multiple human cancers and cells lacking Pokemon are refractory to oncogenic transformation. These findings suggest that Pokemon may offer an effective new target for cancer therapeutics.
POKEMON is a member of the POK (POZ and Krüppel) family of transcriptional repressors, which consists of an NH2-terminal POZ/BTB domain and COOH-terminal Krüppel-type zinc fingers ( Fig. 1 ). The POZ/BTB domain mediates homodimerization and heterodimerization plus recruitment of corepressor/HDAC complexes to these proteins ( 1), whereas the COOH-terminal zinc fingers mediate specific DNA recognition and binding. Currently, >40 POK proteins have been identified in the human genome (http://btb.uhnres.utoronto.ca). Recent reports have uncovered essential roles for POK proteins in development ( 2, 3), differentiation ( 4, 5), and oncogenesis ( 6– 8). For instance, promyelocytic leukemia zinc finger (PLZF)–null mice display severe defects in limb development and germ stem cell maintenance ( 2, 4). Th-POK (T-helper-inducing POZ/Krüppel-like factor, also known as cKrox) has been recently reported as a master regulator of T-cell lineage commitment ( 3). Concerning roles for POK proteins in oncogenesis, B-cell lymphoma 6 (BCL6) and PLZF, have been implicated in the pathogenesis of non-Hodgkin's lymphoma and acute promyelocytic leukemia, respectively ( 5, 6). Moreover, another POK protein family member hypermethylated in cancer-1 (HIC1) is known to be hypermethylated and transcriptionally silent in human cancers. HIC1 heterozygous mice develop spontaneous malignant tumors in multiple tissues ( 7).
POKEMON, which stands for POK erythroid myeloid ontogenic factor (also known as LRF, OCZF, or FBI-1) was originally identified as a protein that binds specifically to a HIV type 1 promoter element ( 9) and was subsequently cloned as a homologue of PLZF that can physically interact with BCL6 ( 10). The Rat homologue was also identified as a key protein involved in osteoclast differentiation ( 11). Pokemon inactivation in mouse resulted in embryonic lethality due to severe anemia and profoundly impaired cellular differentiation in multiple tissues. 1 We hypothesized a potential role for Pokemon in oncogenesis based on its observed role in regulating differentiation, a core facet of a transformed phenotype. To examine this hypothesis, we tested the effect of targeted deletion of Pokemon on oncogenic transformation of primary mouse embryo fibroblasts (MEF). In addition, we examined the expression of POKEMON in a variety of human cancers ( 8).
Using Pokemon wild-type and null MEFs, we tested whether loss of Pokemon function would modulate cell growth or transformation upon oncogenic stimuli. Strikingly, Pokemon null MEFs were almost completely resistant to the effects of all the oncogenic combinations tried, including adenovirus E1A + RasV12, Myc+RasV12, SV40 large T antigen + RasV12 or BCL6 + RasV12. The cells failed to acquire any proliferative advantage and were unable to be transformed (as indicated by colony formation in soft agar). Moreover, MEFs coinfected with Pokemon and the various oncogenes displayed a marked proliferative advantage as well as the ability to form colonies in soft agar. We therefore concluded that Pokemon is critical for oncogenic transformation of MEFs and is able to act as a proto-oncogene in cooperation with other classic oncogenes.
We subsequently attempted to identify Pokemon consensus DNA binding sequences by CAST analysis. This screening selected a specific GC–rich sequence for Pokemon binding, which shows a certain similarity to the consensus sequence for the transcription factor Sp1. An essentially identical DNA binding site was independently characterized by Hernandez et al. using the human POKEMON protein (FBI-1; ref. 12). Because POK proteins are known to recruit corepressor complexes through the POZ domain, thereby acting as transcriptional repressors, we hypothesized that Pokemon might exert its oncogenic activity through the direct repression of potent tumor suppressor or proapoptotic genes. We located several putative Pokemon binding sites in the tumor suppressor ARF promoter and so tested whether Pokemon would repress its activity. Taking advantage of ARF-luciferase reporters and chromatin immunoprecipitation assays, we discovered that Pokemon directly binds the p19Arf promoter in vivo and is able to repress its activity. In the absence of Pokemon, p19Arf expression was found markedly elevated upon both culture shock and oncogenic transformation. Interestingly, expression of the p16Ink4a gene, another tumor suppressor gene encoded in the same Ink4a-Arf locus ( 13), was not elevated in Pokemon-null MEFs, showing the specificity of Pokemon repressive activity. Furthermore, the growth defects and refractoriness to oncogenic transformation in Pokemon-null MEFs were fully reverted by p19Arf loss. These findings showed that Pokemon is a specific repressor of p19Arf and that loss of Pokemon causes aberrant ARF up-regulation, resulting in premature senescence and unresponsiveness to oncogenic stimuli. Interestingly, our preliminary study using Pokemonflox/flox MEFs (established from conditional Pokemon knockout mice) showed that the “acute” loss of Pokemon upon Cre expression also produces proliferative defects and a premature senescence phenotype in MEFs. 2
When exploring the oncogenic role of Pokemon in vivo, we were particularly interested in mouse lymphoma models for the following reasons. First, Pokemon-null embryos show defects in B-cell development. Second, Pokemon is generally expressed in the germinal center of lymphoid tissues and interacts with BCL6, which is also present in germinal center and misexpressed in human B-cell lymphomas. Lastly, preliminary studies indicated that Pokemon is highly expressed in a subset of human T-cell lymphomas, whereas Pokemon is expressed at a low level in CD3-positive T cells of adult thymus. We therefore generated a transgenic mouse model in which Pokemon is overexpressed in immature T and B lymphoid lineage cells using a lckEμ enhancer/promoter transgenic construct. Strikingly, lckEμ-Pokemon mice from two independent transgenic lines developed fatal thymic lymphomas accompanied by lymphadenopathy, splenomegaly, hepatomegaly, and tumor infiltration into bone marrow.
Because Pokemon proved to be an essential factor for oncogenesis both in vitro and in vivo, we went on to investigate POKEMON expression in human cancers by immunohistochemistry using a monoclonal antibody specific for POKEMON. A total of 130 cases of diffuse large B cell lymphomas (DLBCL) and 290 cases of follicular lymphomas were examined for POKEMON protein expression. Significantly, POKEMON was expressed in 60% to 80% of DLBCL and follicular lymphoma cases. Interestingly, POKEMON and BCL6 double-positive cases showed a higher proliferative index (high Ki-67 positivity), suggesting a possible functional crosstalk between POKEMON and BCL6 in lymphomagenesis. Regarding POKEMON expression in T-cell malignancies, our recent analysis of >80 cases of T-cell lymphomas revealed that POKEMON is highly expressed particularly in anaplastic large cell lymphoma and angioimmunoblastic lymphoma cases. 2
The ARF/p53 pathway is frequently abrogated in human cancers. However, whereas mutations in the ARF/p53 pathway are relatively rare in DLBCL compared with solid tumors ( 14), both Arf and p53 mutant mice commonly develop lymphomas ( 15, 16). Our real-time reverse transcription-PCR analysis of a cohort of DLBCL cases revealed that high Pokemon gene expression generally correlated with low expression of the p14ARF gene, which underscores the potential importance of ARF suppression by Pokemon in DLBCL. Given that Pokemon is also overexpressed in solid tumors such as colon cancer and bladder cancer in which the normal function of the ARF/p53 pathway is frequently lost, it is likely that Pokemon has multiple additional target genes by which it can exert its oncogenic activity. In this respect, it is worth noting that Kaiso, another POK family member, specifically binds to methylated DNA sequences and further enhances repression of target genes ( 17). It is tempting to speculate that Pokemon also exerts its oncogenic function through epigenetic control of the target gene (i.e., promoter hypermethylation). Furthermore, recent reports have uncovered a functional correlation between POZ/BTB domain–containing proteins and Cullin-based E3 ligases. POZ domain protein family acts as substrate-specific adaptors of Cullin3-based E3 ubiquitin ligase complexes and plays a role in substrate targeting of cullin-based E3 ligase complexes ( 18). Pokemon may therefore be able to regulate targets or pathways through protein ubiquitination. Whereas >40 POK proteins have been identified in human, little is known about the “functional crosstalk” among these proteins. Given that POK proteins can form heterodimers/homodimers through the POZ/BTB domain (e.g., PLZF/BCL6 and Pokemon/BCL6) and these proteins could share similar DNA consensus sequences (e.g., PLZF and FAZF; ref. 19), functional networks that are governed by POK family proteins are likely more dynamic and complicated than originally anticipated. Furthermore, it is interesting to hypothesize possible interactions/crosstalk between POK proteins and other POZ/BTB containing protein families, which contain >100 proteins in human. The POZ/BTB domain is normally found as a single copy in proteins that also contain one or two other domain types such as Zinc finger (POK proteins), BACK-kelch, voltage-gated potassium channel T1, and MATH domains ( 20). The multitude of potential interaction partners for POK proteins would be expected to lead to a great diversity in their cellular functions beyond simple target gene repression.
Although we found Pokemon to be aberrantly overexpressed in human cancer, little is known about the mechanism by which it becomes overexpressed in cancer. Thus far, genetic evidence directly linking POKEMON with human cancer is lacking (e.g., chromosomal translocations and mutations of the POKEMON locus). However, the genomic locus where the POKEMON gene resides (chromosome 19p13.3) is a gene-rich region that is found frequently mutated in human cancers. Interestingly, one recent study showed that the t(14;19)(q32;p13.3) translocation is one of the more common cryptic translocations involving the immunoglobulin heavy chain gene in B-cell non-Hodgkin's lymphoma ( 21). It is also possible that POKEMON up-regulation would be due in some instances to the aberrant activation of upstream regulatory pathways. In this respect, Astier et al. recently reported that POKEMON (FBI-1) is induced upon fibronectin-mediated β1-integrin ligation in precursor B leukemia cells ( 22). The authors speculate that up-regulation of Pokemon facilitates cellular proliferation upon fibronectin binding. Further studies will be necessary to precisely determine how POKEMON expression is regulated both in normal and cancerous cells.
In closing, our findings suggest that POKEMON could be an attractive therapeutic target for human cancer therapy in view of its essential role in oncogenic transformation. Indeed, the possibility of a strategy to elicit functional blockade of POKEMON has been proposed recently by A. Melnick et al. ( 23). This group developed a blocking peptide for the POK family protein BCL6 that specifically binds the “lateral groove” of its POZ domain and abrogates corepressor interaction, resulting in the effective blockade of BCL6 function both in vitro and in vivo. A similar strategy may also be useful for the knockdown of POKEMON function in cancer cells. Indeed, we are now testing whether inhibition of Pokemon function could lead to the eradication or prevention of malignant transformation and progression both in vitro and in vivo using a conditional Pokemon mouse knockout model.
↵1 Merghoub and Pandolfi, unpublished observations.
↵2 Maeda and Pandolfi, unpublished observations.
- Received March 29, 2005.
- Revision received July 18, 2005.
- Accepted July 19, 2005.
- ©2005 American Association for Cancer Research.